On Tue, Oct 18, 2022 at 6:04 PM David G. Johnston <david.g.johnston@xxxxxxxxx> wrote: > On Tue, Oct 18, 2022 at 8:53 AM Dominique Devienne <ddevienne@xxxxxxxxx> wrote: >> I'm surprised by the result for bit(3) and char, when calling pg_column_size(). > The base type is what matters, if the length of the actual type is a parameter > (the (n) part) the underlying type must be variable. Thanks. Interesting. Didn't know (n)-suffixed "fixed-length" types where always based on variable-size ones. >> How does one store as compactedly as possible several small enums > int2 OK, I see. Thanks again. > p.s., pretend char doesn't even exist. I realize that now. Wasn't obvious to me, despite the warning in the doc.